An interframe prediction technology in the High Efficiency Video Coding (HEVC) adopts a conventional Motion Compensated Prediction (MCP) method. For motion vector prediction, the HEVC adopts a multiple motion vectors competition method, which improves the accuracy of motion vector prediction, thereby improving the coding compression performance.
The HEVC interframe prediction mode may include, but is not limited to, a Merge mode, a Skip mode, and so on, all of which use the multiple motion vectors competition to perform interframe prediction. A motion vector list is used when motion vector prediction is performed. For the Merge mode and the Skip mode, the motion vector list is allowed to contain at most four spatial motion vector predictors and one temporal motion vector predictor, where the Merge mode and the Skip mode share one motion vector list. A coder selects a best motion vector predictor from the motion vector list as a motion vector predictor of a current Prediction Unit (PU), which may be called a prediction block hereinafter.
A method for building the motion vector list may include:
As shown in FIG. 1, spatial neighboring blocks of the current PU may include: a neighboring block A0 (a bottom-left reference block corresponding to a bottom-left corner position of the current PU), a neighboring block A1 (a left reference block corresponding to the bottom-left corner position of the current PU), a neighboring block B0 (a top-right reference block corresponding to a top-right corner position of the current PU), a neighboring block B1 (a top reference block corresponding to the top-right corner position of the current PU), and a neighboring block B2 (a top-left reference block corresponding to a top-left position of the current PU). A Temporal Motion Vector predictor (TMVP) is a motion vector corresponding to the current PU in time domain.
Firstly, motion vectors are successively obtained from the neighboring blocks and the TMVP in a sequence of the neighboring block A1, the neighboring block B1, the neighboring block B0, the neighboring block A0, the neighboring block B2, and the TMVP. Then, the obtained motion vectors as motion vector predictors are added to the motion vector list according to a rule in the HEVC. The specific motion vector list building process is a technology well known by a person skilled in the art, so that details will not be described herein again.
Further, a current Coding Unit (CU), which may be called a coding unit hereinafter, may include at least two PUs, and motion vector lists of the at least two PUs are built in a serial manner.
During the implementation of the above motion vector list building process, it is found that the prior art has at least the following problem: if the current CU includes at least two PUs, and the motion vector lists of the at least two PUs are built in a serial manner, the speed of building the motion vector lists of the at least two PUs in the same CU is slow, thereby reducing the parallel processing capability.